DI

SC

US

SI

ON

P

AP

ER

S

ER

IE

S

Forschungsinstitut zur Zukunft der ArbeitInstitute for the Study of Labor

Skill Variety, Innovation and New Business Formation

IZA DP No. 7889

January 2014

Jolanda HesselsUdo BrixyWim NaudéThomas Gries

Skill Variety, Innovation and

New Business Formation

Jolanda Hessels ESE, Erasmus University Rotterdam and Panteia/EIM

Udo Brixy

Institute for Employment Research (IAB) and Ludwig-Maximilians University, Munich

Wim Naudé

MsM, University of Maastricht and IZA

Thomas Gries University of Paderborn

Discussion Paper No. 7889 January 2014

IZA

P.O. Box 7240 53072 Bonn

Germany

Phone: +49-228-3894-0 Fax: +49-228-3894-180

E-mail: iza@iza.org

Any opinions expressed here are those of the author(s) and not those of IZA. Research published in this series may include views on policy, but the institute itself takes no institutional policy positions. The IZA research network is committed to the IZA Guiding Principles of Research Integrity. The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center and a place of communication between science, politics and business. IZA is an independent nonprofit organization supported by Deutsche Post Foundation. The center is associated with the University of Bonn and offers a stimulating research environment through its international network, workshops and conferences, data service, project support, research visits and doctoral program. IZA engages in (i) original and internationally competitive research in all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research results and concepts to the interested public. IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.

IZA Discussion Paper No. 7889 January 2014

ABSTRACT

Skill Variety, Innovation and New Business Formation We extend Lazear’s theory of skills variety and entrepreneurship in three directions. First, we provide a theoretical framework linking new business creation with an entrepreneur’s skill variety. Second, in this model we allow for both generalists and specialists to possess skill variety. Third, we test our model empirically using data from Germany and the Netherlands. Individuals with more varied work experience seems indeed more likely to successfully start up a new business and being a generalist does not seem to be important in this regard. Finally, we find that innovation positively moderates the relationship between having varied experiences, and being successful in starting up a new business. Our conclusion is that entrepreneurs with more varied work experience are more likely to introduce innovations that have not only technical, but also commercial value. Our findings support the notion that entrepreneurship can be learned. JEL Classification: L26, M13, J24, O31 Keywords: entrepreneurship, start-ups, human capital, innovation, skills Corresponding author: Wim Naudé Maastricht School of Management P.O. Box 616 6200 MD Maastricht The Netherlands E-mail: naude@msm.nl

1 Introduction

The creation of a new business can contribute to development by fuellingcompetition, creating jobs, and providing new goods and services (Acs andNaude, 2013; Braunerhjelm et al., 2010). These are amongst the societalcontribution of entrepreneurs, who identify and utilize risky opportunitiesthrough the creation of a new business (Gries and Naude, 2011). Creatinga business entails a long list of tasks. The prospective entrepreneur needsto identify an opportunity, conceptualize a business plan, organize financ-ing, location, production techniques and also market of the new business,amongst other tasks.

This variety of tasks requires much skill, hence it is not surprising thatmany nascent entrepreneurs fail. As a result when governments or devel-opment agencies want to stimulate new business start-ups, they most oftenchoose to support initiatives that aim to improve the skills of prospectiveentrepreneurs (Bruhn and Zia, 2011; Karlan and Valdivia, 2011). Scholarlyresearch indeed confirms that skills is an important determinant of sucess-full entrepreneurship (see e.g. Astebro and Bernhardt, 2005; Baptista et al.2007; Evans and Jovanovic, 1989; Kihlstrom and Laffont, 1979; Lucas 1978).As put by Kanniainen and Poutvaara (2007:676) ‘people differ substantiallyin terms of their ability to produce a business idea, elaborate their idea, andmake its way to a marketable product or service’. In a nutshell, a successfulnew business is the result of a ‘match’ between the skills of a prospectiveentrepreneur and opportunities.

There is a growing literature on what kind of skills an entrepreneur needs.According to Lazear (2004, 2005) an entrepreneur needs to be a ‘jack-of-all-trades’. His argument is that the creation of a new business requires manydifferent activities to be performed, as mentioned, so that entrepreneursneed to be multi-skilled. Also, having a variety of skills may help to suc-ceed in bringing an innovation to the market, or to commercialize a newidea. Generalists are, in other words, better entrepreneurs than specialistsin Lazear’s view.

The contribution of our paper is to extend Lazear’s theory in three direc-tions. First we provide, for the first time as we are aware, a theoreticalframework to link entrepreneurial skills to opportunities. In this model weillustrate that it is the interaction of opportunities, skills and innovationthat result in the creation of new businesses. Second, we refine and extend

1

Lazear’s (2005) ‘jack-of-all-trades’ notion to allow for both generalists andspecialists to exhibit a variety of skills as a result of their experiences and/orinvestments in human capital. In other words in our model entrepreneurscan have a variety of skills, even if they specialize. Our third contribution isto test our model empirically using data from Germany and the Netherlands.

We find that individuals with more varied work experience are more likelyto succeed in their attempts to start a new business. Moreover being a gen-eralist, which proxies having balanced skills, does not matter for success instarting a new business. We also establish that innovation positively moder-ates the relationship between having varied experience and being successfulin setting up a business. Entrepreneurs with more varied work experienceare more likely to propose innovations that not only have technical, but alsobusiness value. These findings support the notion that entrepreneurship canbe learned and that a variety of experience matters, not only for the per-formance of existing firms (as in Toft-Kehler et al., 2013) but also for thesuccess of new business start-ups.

The rest of the paper proceed as follows. In section 2 we discuss the extantliterature on skills and entrepreneurship. In section 3 we propose a noveltheoretical model to link variety of skills, opportunities, and business cre-ation. We derive a number of testable hypotheses in section 4. Then, insection 5 we use a new dataset from Germany and the Netherlands to testthese hypotheses. Section 6 concludes.

2 Literature Review

One of the most influential theories in the context of entrepreneurial abil-ity is the theory of balanced skills, or ‘jack-of-all-trades’ theory of Lazear(Lazear 2004, 2005). Lazear is concerned with the individual choice be-tween paid-employment and self-employment. In his model individuals areendowed with two skills: an individual can be a specialist, in which casehe or she receives income associated with his or her best skill, or can bean entrepreneur, in which case he or she is limited by his or her weakestattribute (Lazear 2005: 652). This would imply that individuals who areespecially strong in one field would tend to invest even more in this line ofcompetence, so as to obtain the maximum benefit as a highly specializedpaid employee. On the other hand those who are multi-skilled with no com-petence that especially stands out, would in contrast be relatively better off

2

in self-employment.

Several studies have empirically tested this idea including Lazear (2004,2005), Wagner (2003, 2006) and Silva (2007). Lazear (2004, 2005) usesStanford alumni data (Graduate School of Business Alumni, late 1990s)and finds that those who at some point became entrepreneurs had a morevaried curriculum. From this he concludes that entrepreneurs adopt a moregeneral human capital investment profile before they enter the labor mar-ket. Furthermore he found that the likelihood of becoming an entrepreneuris higher when someone has taken on a larger number of roles with previ-ous employers, which he derives from the number of different occupationaltitles one has had (Lazear, 2005). Hence those who eventually become en-trepreneurs have apparently benefitted from having taken on a variety ofdifferent roles since entering the labor market. The latter may have beenthe result of differences in individual’s endowments of general skills, or mayhave been due to individual’s conscious human capital investments. Theinvestment view dominates, according to Lazear (2005).

The idea that entrepreneurs are ‘jacks-of-all-trades’ has also been estab-lished in studies using cross-sectional data from Germany (Wagner 2003;2006; Bublitz and Noseleit, 2013) and Canada (Astebro and Thompson,2011). Silva (2007) however, using panel data from Italy and controlling forunobserved heterogeneity finds no support for this theory.

Earlier studies testing the ‘jacks-of-all-trades’ hypothesis typically comparethe skills of entrepreneurs with those of wage-employees. The idea thanis that having a variety of skills helps to keep a business up and runningand to survive. It could also be argued that having a variety of skills helpssomeone to get a business started when in the process of setting up a firm.This is a phase in which prospective entrepreneurs need to combine differ-ent resources (e.g. financing, technology and labor). Instead of answeringthe question why some choose to become entrepreneurs rather than wage-employees, in this paper we answer the question why some entrepreneurssucceed in establishing a business, when many try but do not succeed.

Lazear (2005) considers those who initially established the business as en-trepreneurs as these individuals have usually been responsible for the con-ception of the basic product, hiring the initial team, and obtaining at leastsome early financing. Thus, the acts or transactions needed to establisha business and get a business up and running are seen as entrepreneurial.Spulber (2009) goes even further by defining an entrepreneur as someone

3

who establishes a business, which is what makes them different from othereconomic actors. He distinguishes between the decision to become an en-trepreneur and the actual foundation of the business. Being an entrepreneurlasts from the decision to set up a business to the moment that the businessis actually founded. During this period the entrepreneur devotes resourcesand efforts to establishing the business. Once an individual succeeds in es-tablishing a firm he is no longer an entrepreneur but the owner of the firm.The established firm is now a new independent economic actor with ownobjectives distinct from those of the owner, while the objectives of the en-trepreneur are interconnected with those of the start-up during the phaseof firm foundation. We focus on whether those entrepreneurs who made adecision to start up a new business are more likely to succeed when they aremultiple skilled.

Success in getting a new business founded is by no means guaranteed. Start-up attempts may turn out to be unfeasible and may be abandoned, even ifthey were planned in advance (Lazear, 2005). This selection during theprocess of starting-up a firm is by no means needless as it helps to singleout projects that are promising. A high proportion of business fail in thefirst months after start-up, sometimes described as due to the ‘liability ofnewness’ (Freeman et al. 1983, Bruderl and Schussler 1990). These highrates of firm closures show that this selection actually still is not efficientenough.

Two earlier studies have investigated whether the number of fields in whichan individual has experience matter for start-up success. These measuredsuccess according to the number of gestation activities completed (Stuetzeret al., 2012 and Stuetzer et al., 2013a;2013b). They found that entrepreneurswith a broad set of skills (which they term ‘skill balance’) complete moregestation activities. Toft-Kehler et al. (2013) investigated the impact ofexperience on the financial performance of existing firms using Swedish data,and found that the firms headed by owners/ managers with more variedexperience tend to perform better. It remains unclear, however, whetherskills and variety of experiences contributes to start-up success.

Lazear (2005) argued that it is not the sheer number of skills mastered alonethat makes people prefer to become self-employed or not, but that it alsodepends on the balance of these skills. There is some tension in this ideabecause it is clear that entrepreneurs will often need to be specialists whenthe demand for more specialist goods are higher, and /or where high ratesof innovation are required to enter markets (so that unique goods or services

4

can be developed that needs specialist knowledge, e.g. in information andcommunication technology).

While Lazear (2004, 2005) equates mastering skills in various areas withbeing a generalist we deviate from this by proposing that those who mastervarious skills can still be specialists (and that these entrepreneurs could infact be more innovative than specialists without varied skills or generalistentrepreneurs without specialist knowledge). The mastering of various skillscan result from a deliberate human capital investment strategy based onthe recognition that certain skills are required for being an entrepreneur,whether one is a specialist or more of a generalist. Therefore we propose tomake a distinction between the amounts of variety in work experience (skillvariety), and whether someone is more of a generalist or more of a specialist(skill balance). While the first captures variety in skills the later proxiesthe extent of balance of the skills one possesses. One can have a varietyof skills but still be a specialist, which is not accounted for in Lazear’smodel (Lazear, 2005). We allow for individuals to both invest in a varietyof skills and still be a specialist as they mainly excel in one skill. While aspecialist in Lazear’s model only invests in one of the two skills, but not inboth, in reality individuals may invest in more than one skill, while theycan still be mainly specialists. We propose that in particular skill varietywill contribute to success in establishing a business. A varied or broad skillset has been associated with higher entrepreneurial skills relevant for settingup a business (Stuetzer et al., 2013b) and with completing a larger numberof gestation activities during the period when one attempts to establish abusiness (Stuetzer, et al.,2012).

Lazear (2005) further argues that multi-skilled entrepreneurs may have anadvantage when it comes to introducing an innovation to the market becauseit is easier to innovate “when the entire situation can be seen” (Lazear, 2005:661). Thus, specialists may be at a competitive disadvantage to those thathave a variety of skills in particular when attempting to bring a new in-novation to the market. While a technical specialist may be excellent atdeveloping some new device, this does by no means guarantee that thisdevice will be commercially viable. To be successful as an innovative en-trepreneur, business relevance is needed; technical soundness alone is notenough (Lazear, 2005: 661). Those with a varied skillset are likely to bebetter in developing new products or services that have actual business valueand therefore skill variety may help to succeed in bringing an innovation tothe market, or to commercialize a new idea. Lazear (2005) further arguesthat this may apply to product innovation but also to the use of business

5

process innovation. Entrepreneurs with skill variety may be better at usinga new business process that allows firms to beat the competition. To testthis idea that innovation and skill variety matter we also attempt to capturein this paper the interaction between an individual’s experience and skillson the one hand, and the attempt to innovate by introducing a new productand/or service or by introducing a business process innovation. As Lazear(2005) did not test this prediction a contribution of this paper is to do so.Before doing so, we propose in the next section a theoretical framework tobetter understand the relationship between skills variety and entrepreneurialstart-up outcomes.

3 Modelling Skills Variety and Balance and NewBusiness Creation

As mentioned, entrepreneurs are individuals who spot and use opportuni-ties to create new businesses. Their success depends on their skills and theavailability of opportunities that matches their skills. Lazear (2005) arguedthat those with more generalist skills will fare better than those with morespecialised skills in creating a new business firm. In this section we formalizethis idea by linking skills variety and entrepreneurship with opportunities.Our basic idea of linking skills variety with new business creation throughthe identification and use of opportunities is based on labour economics’matching models. In these models, various ‘frictions’ prevent a matching ofvacancies with labourers’ skills giving rise to unemployment. We use thisidea to describe the skills profile of latent and nascent entrepreneurs as de-termining the extent of their effort to search for opportunities and overcomefrictions in their utilization. As stated previously, our key hypothesis, fol-lowing Lazear (2005), is that entrepreneurs with a variety of skills are morelikely to create new businesses - in our model because such entrepreneursare more likely to find a “match” with an opportunity. To model the match-ing of opportunities and business ideas we use basic concepts of matchingtheory as suggested e.g. by Pissaridis (2000) for labor markets and transferthem to the entrepreneurial market entry decision, following first steps inthis direction by Gries and Naude (2011).

We present our model in five sub-sections or stages. First, we describethe pool of entrepreneurial talent in our economy, distinguishing betweenlatent, nascent and existing entrepreneurs. Second, we introduce the setof opportunities for creating new businesses, making a distinction between

6

opportunities that require innovation, and more informal, less productiveopportunities. Third, we explain how the former concepts of entrepreneursand opportunities can be linked to result in new business creation as amatch between the set of entrepreneurial skills and opportunities. Fourth,we elaborate on the distribution of various skills and business success. Fifth,we derive the equilibrium outcome in our hypothetical economy, from whichtwo testable hypotheses are generated.

3.1 Entrepreneurs

Most new businesses are created by entrepreneurs. Within our model wedistinguish between active entrepreneurs n and nascent entrepreneurs u.In contrast to the active entrepreneur the nascent entrepreneur is in thebeginning stages of establishing a business, and may still be in a phaseof evaluating an opportunity. Nascent entrepreneurs are defined accordingto the GEM definition as those who are actively taking steps to set upa new business (for a detailed definition including precise questions askedsee Parker (2009: 8)). The concept of nascent entrepreneurship has theadvantage that it is neither prone to a survival bias nor to a hindsight bias.According to Parker (2009), however, a major disadvantage is that not allstart-ups are preceded by a period of planning, but are rather spontaneouslyfounded. This would imply a certain underestimation of the number of start-ups. This could be offset by the fact that only a part of all planned start-ups,however, are eventually started. It is difficult to define a precise moment forwhen exactly someone has become a nascent entrepreneur. Often there isno exact point in time at which someone becomes a nascent entrepreneur –rather it is a gradual process where one step follows another. Furthermore,it is difficult to observe the exact moment of a firm’s inception or successin setting up a business (Katz and Gartner, 1988; Reynolds and Miller,1992; Carter, Gartner and Reynolds 2004). Here inception is based onself-assessment of the nascent entrepreneur on whether the new venture isoperational or not.

Having made the distinction between nascent and active entrepreneurs, itfollows that the total entrepreneurial potential in the economy, E, is the sumof active and nascent entrepreneurs, which can be written as E = n+ u.

7

3.2 Opportunities

In order to create a new business, nascent entrepreneurs evaluate their start-up opportunities. Opportunities are exogenously given and exist indepen-dently of entrepreneurs. Ω represents the total number of potential newbusiness opportunities in the economy. These opportunities can be dividedinto three different groups. (i) Opportunities already taken, which lead tothe total of n active entrepreneurs and their start-up firms, (ii) a numberω of unrealized profitable opportunities ready for being taken by a nascententrepreneur on the lookout, and (iii) there are unrealized but idle (or yetunproductive) opportunities available, denoted by δ.

These could be temporary or informal opportunities which are at presentnot lucrative, such like self-employment and survivalist-type activities. Thusthe total number of start-up-opportunities can be written as Ω = n+ω+ δ.

3.3 New Business Creation as a Match

At the time when the nascent entrepreneur spots and then utilizes an op-portunity that matches their skills a business is founded. The number ofnew firms that start as the result of such matching per period is denoted byM. In aggregate this matching rate - or new business creation rate - will bedetermined by three factors: (i) the specific conditions in a particular mar-ket segment (including the business environment) that affect the efficiencyµ of a matching process between a product profile and opportunities; (ii)the extent of unrealized profitable opportunities, ω in relation to the num-ber of nascent entrepreneurs trying to spot these opportunities, and (iii) thenascent entrepreneurs’ skills variety that positively affects the search anddevelopment effort for finding opportunities η.

The aggregate effective search and development effort as a function of skillsvariety is hence ηu. Depending on the extent and variety of skills the ef-fort of nascent entrepreneurs may vary considerably as frequently stated inthe related literature. An important point is that the type of skills (skillsvariety) interacts with the local environment or specific market conditionsso that the new business creation rate in a country is not only the result ofentrepreneurial skills, or only the result of the business environment - bothmatter, and in fact the business environment and specific market conditionsmay determine the nature of opportunities which again will determine which

8

skills will be better rewarded. A practical policy implication that followsis that entrepreneurial education and business environment reform need tobe tackled jointly and in coordinated fashion to result in more and bettermatches between skills and opportunities.

A theoretical implication that follows is that entrepreneurs with a variety ofskills may also specialize due to the demands of the environment / nature ofopportunities that they spot. For instance, if nascent entrepreneurs are in amarket that demands more specialized and differentiated products, this mayencourage innovation and specialization. Hence market matching efficiencyµ is affected by the degree of specialization in a market. More specializedgoods have a more precise profile and nascent entrepreneurs find a matchmore easily. In a specialized market an innovative product variation willbe easily identified and such an innovation will more easily find customerswho benefit from the innovative idea. In the next section we will discussin more detail the endogenous decision for search intensity and relate thisdecision to the entrepreneur’s skills variety and search and information costsas well as to the matching efficiency of a specific market. After havingintroduced the determinants of the matching process the matching (andhence start-up) rate, M , can be specified. Throughout the paper we willassume linear homogeneity of the matching function. Hence the effectivenessof the matching process is not a matter of the size of the economy, butrather of market specific conditions µ combined with the variety of skillsthat determines search effort and the start-up rate.

For computational simplicity we will model the matching-function as aCobb-Douglas function M = µM = µωβ (ηu)1−β. Hence, the probability ofa successful new firm start-up is µM

u = µm.

3.4 Optimal Search and Survival Effort, Matching, and FirmFailure

We can now bring the previous three building blocks together to link newbusiness creation with skills through the search and development effort it en-ables. At the individual level, any nascent entrepreneur i will have to makesome search and development efforts described by the effort ηi to identifyand develop a new business opportunity. As mentioned the collecting ofinformation and searching for opportunities is costly. The search and de-velopment cost per unit of effort of the nascent entrepreneur is denoted ci.

9

In order to safeguard their firms’ survival, active (existing) entrepreneurswill have to invest a certain survival effort, ψi. The optimal search and de-velopment effort to enter the market and the optimal survival effort to stay inthe market, results from the maximization of the entrepreneurs’ net presentvalue11. Hence we have (i) nascent entrepreneurs searching and implementingopportunities to start a new business, and (ii) active entrepreneurs tryingto stay in business (be successful).

i) Nascent entrepreneurs: A nascent entrepreneur is in the early stages ofthinking about starting a business and evaluating opportunities to start-upa firm. Net present value of all activities of this state is Wi and containscurrent wage income wi minus search costs ci multiplied by the search ef-fort ηi plus the expected extra entrepreneurial income in case a prosperousopportunity is spotted and realized as a lucrative new start-up enterprise.

In order to turn to a more detailed modeling of the nascent entrepreneurs’search and development costs we now modify the entrepreneurial matchingmodel introduced by Gries and Naude (2011) and apply it to the describedproblem. As a first step we argue that skill variety determines search costs.Marginal search costs ci are a function of some given costs which we normal-ize to one and the variety of experiences. The idea is that the variety of workexperiences gives insights into the functioning and thinking of the potentialcostumer side. If a nascent entrepreneur worked at many firms or positionsbefore he or she will be able to more easily address the relevant items andhence lower his or her search costs. Similar, if a nascent entrepreneur is e.g.an engineer (with originally human capital in engineering) that has to placehis product in the market (requiring a marketing skill), finance the start-uploans (requiring financial knowledge), be informed about legal and labor lawrestrictions (have knowledge on law issues), pay taxes (have knowledge ontax laws) and manage his firm organization (have organization skills andmethods) he or she must be able to handle many very different tasks onhis own with low learning costs. Only when his or her firm is growing heor she may be able to divide these specific fields of activities and allocatespecialists to them. Hence the variety of work experiences can be expectedto clearly matter for the costs of finding and implementing opportunities

1 We can also introduce unemployed persons searching for opportunities while still onwelfare benefits, but for the sake of tractability we abstract from this possibility fornow.

10

and the success of a start-up. In our model, each field j of experiences saveslearning costs and adds as a costs reducing element of entrepreneurial hu-man capital to individual search productivity with the effect of hj . Hencetotal marginal search and development costs are a reverse of entrepreneurialhuman capital consisting of the number of work experiences of entrepreneuri and hence of the nascent entrepreneurs skill variety ei:

ci = (

ei∑j=1

hij)−1

The larger the number of different work experiences (the larger the vari-ety) the more productive is each unit of search effort and the lower arethe marginal costs of search activities. The variety of abilities matters asthe nascent entrepreneur is familiar with more components of the businessactivity.

Further, denoting Vi as the expected value of entrepreneurial income, thenthe general expected extra entrepreneurial income ∆i can be written as

∆i = Vi −Wi.

However, being a nascent entrepreneur this extra income is only realizedwhen finding the opportunity to start-up. Hence for a nascent entrepreneurthe expected extra income is ∆i times the probability of matching and find-ing an opportunity. As argued in the previous section, this probability isµmi. As individual efforts determine the matching probability mi (ηi) for agiven discount rate r we obtain

rWi = wi − (

ei∑j=1

hij)−1ηi + µmi (ηi) ∆i

ii) Active entrepreneurs: As an active entrepreneur is continuously work-ing to keep his or her firm going, the net present value of being such anentrepreneur Vi is

rVi = vi − γiψi − φi (ψi) ∆

In this second state the profit flow is vi. In order to stay in the marketthe entrepreneur decides to invest γiψi with effort ψi and marginal costs

11

γi. The transitory and dynamic character of markets and role of the en-trepreneurs abilities for founding and running the company are reflected bythese required investments and efforts. However effort towards firm survivaland investments do not guarantee that a firm will not fail and entrepreneurshave to consider the likelihood of firm failure denoted by φi. Taking theperspective of an individual entrepreneur i however reveals that her sur-vival efforts ψi may reduce the likelihood of firm failure φi which follows

φi = φi(ψi), φψi:= ∂φi

∂ψi< 0, φψiψi

:= ∂2φi∂ψ2

i> 0.

Hence, an entrepreneur having a particular skill variety and experience hasthe choice to invest with search effort ηi and with survival effort ψi in orderto increase the chance of finding a match and to lower the probability of afirm failure. No matter what state an entrepreneur is in, she can maximizethe expected income. Thus the following maximization exercise will give usthe optimal search effort and the survival effort:

maxηi

: rWi = wi − ciηi + µmi (ηi) ∆

maxψi

: rVi = vi − γiψi − φi (ψi) ∆

We obtain from the first order conditions (F.O.C) an optimal search effortη∗ and optimal survival effort ψ∗ by using the implicit function theorem22

η∗i = η∗i (u, ω,∆, ei, hij , µ) , withηu < 0, ηω > 0, η∆ > 0,ηei > 0, ηhi > 0 ηµ > 0

ψ∗i = ψ∗i (∆, γi) , with ψγi < 0, ψ∆ > 0

Most interesting, partial derivatives ηei > 0 and ηµ > 0 suggests that searcheffort will increase with a larger skill variety ei and with a higher matchingefficiency µ which may be due to more specialized and innovation drivenmarkets.

3.5 Aggregate Equilibrium Outcome

As we assumed symmetry across all entrepreneurs, we can now focus on theaggregate equilibrium. We start with the representative wealth differential

2 See appendix 1.

12

∆ of both wealth levels (W and V ) corresponding to being a nascent en-trepreneur (in the process of evaluating an opportunity), or being an activeentrepreneur (trying to stay in business). We obtain an implicit relation forthis wealth differential by

∆ =v − w + cη (x)− γψ (∆, γ)

r + φ (ψ (∆, γ)) + µm (η (x))(1)

with x defining the vector x = (u, ω,∆, e, h, µ).

Next, we calculate the difference in new start-ups and firm failures that willdescribe the market dynamics. From the above we know that on aggregatethe number of firm failures is φn, and the total of matched new businessstart-ups equals µM . Hence net creation of firms is n = µM − φn. Thecorresponding stationary flow equilibrium condition is:

n = 0 ⇔ M = µM = φn (2)

In stationary equilibrium the number of new firms will equal the number offirm failures.

Eventually we want to determine the aggregate stationary number of busi-nesses. Hence, we have to account for the dynamics of opportunities in theeconomy. In this model this dynamics are covered by two probabilities de-noted p and q. p is the probability that profitable opportunities – no matterif actively used or unused – turn unprofitable, whereas q denotes the prob-ability of once idle opportunities becoming lucrative. p and q may dependon exogenous factors such as having a boom or bust during a business cyclewhich we do not consider here. Hence the dynamics (rate of change) inidle start-up opportunities is δ = p (ω + n) − qδ, and the stationary flowequilibrium for opportunities is33

δ = 0 ⇔ ω =q

p+ qΩ− E + u (3)

The equations (11), (22) and (33) form a system with three endogenous variables(u, ω,∆). The considered determining factors in this system are skill varietye and market specific conditions represented by the matching efficiency µ.

3 We use the definition Ω = n+ ω + δ to substitute for δ.

13

Both factors affect search effort η and eventually the aggregate start-up ratein this economy.

0 = F = φ (ψ∗) (E − u)− µM(ω, u, η∗) stationary matching equilibrium

0 = G = ∆(r + φ (ψ∗) +µM(ω, u, η∗)

u)− v + γψ∗ + w − cη∗ wealth diff.

0 = H = ω − q

p+ qΩ + E − u supply of profitable opportunities

From this system of equations we can derive Proposition 1.

Proposition 1: The economy [the system of equations F , G, H ] hasa stationary matching equilibrium solution of new business creationand business failure, and hence a stationary number of nascent en-trepreneurs u∗, unrealized but profitable opportunities ω∗ and a sta-tionary differential of entrepreneurial and labor wealth ∆∗, as long asqp+qΩ− E > 0→ ω > u .

u∗ = u∗ (x) , ω∗ = ω∗ (x) , ∆∗ = ∆∗ (x) ,

where x = (µ, e, h) represents the vector of considered exogenous vari-ables.

Proof: See Appendix 2.

Proposition 1 means that there is a stable number of businesses in the econ-omy in equilibrium. Being confronted with a constant number of firms wecan investigate to what degree opportunities or entrepreneurial potential inthis economy could be utilized. The fact that we can determine the station-ary wealth premium ∆∗ for a representative entrepreneur is another featureof Proposition 1. Knowing this information means that one can deduce theeconomy’s efficiency with respect to entrepreneurial activities. Assuming aperfect and risk neutral market economy there is little need for an extrapremium to become an entrepreneur. Therefore, we describe the market asa location (or institutional framework) which may or may not be fulfillingits purpose efficiently.

14

3.6 Hypotheses and Comparative Statics

So far the model argued that skill variety and various work experiences mayaffects information, transaction and search costs that are a major element tofind opportunities and to start up a firm. We can now show that the modelimplies that an increase in skill variety matters for success and increase thestart-up rate (as defined here as the percentage rate of new firms in relationto existing firms).

Hypothesis 1: An increasing variety of work experiences, de > 0, willreduce information and search costs and make nascent entrepreneursmore confident to succeed, leading to (i) an increasing start-up effortdη > 0, and (ii) an increase in the matching and the equilibrium start-up rate ε = M

n , as long as qp+qΩ− E > 0→ ω > u,

dη

de> 0,

dε

de> 0

Proof: See Appendix 3.

As a result entrepreneurs with a variety of work experiences should be moresuccessful in the start-up process.

A second hypothesis can be also derived from our model. If nascent en-trepreneurs are in a market with more specialized goods demand and supplyprofiles become more precise. Entrepreneurs developing an innovative prod-uct even may have clear ideas of specific costumers and their needs. Thatis, in contrast to a product development of a rather unspecified product foran anonymous market new innovative entrepreneurs in specialized marketshave a clear idea of costumers and their specific needs. They will be ableto introduce a specific innovative product variation to the particular targetgroup. For example if an engineer develops a new electronic component fora niche market she will be able to give a very precise description of theproduct innovation. She may even have developed the innovative productfor a group of potential costumers she had in mind; and she may have takencare of specified elements already during the R&D process. Hence, underthis market conditions the product profile of the innovation can be moreeasily compared with a product profile a costumer would like to have, such

15

that profiles easily match or not. As a result the efficiency of the match-ing process is comparably high. Markets with specialized products have ahigher matching efficiency when innovations appear and hence the matchingefficiency increases in such markets. Therefore this discussion results in thefollowing testable hypothesis:

Hypothesis 2: If in specialized markets innovations and innovative el-ements can be easily identified such that matching efficiency rises,dµ > 0, and as a result, (i) start-up effort of innovative nascent en-trepreneurs η increase, and (ii) the matching and start-up rate ε = M

nimproves

∂η

∂µ> 0

∂ε

∂µ> 0.

Proof: See Appendix 3.

In the next section we attempt to confront these hypotheses with empiricaldata.

4 Empirical Study

4.1 Data and Key Variables

In the previous section we derived two testable hypotheses. In this sectionwe try to test these. First however a word about the data and key variables.

4.1.1 Nascent entrepreneurs

Nascent entrepreneurs (those in the process of starting up a business) wereidentified from the adult population survey of the Global EntrepreneurshipMonitor (GEM) conducted in spring 2006 and spring 2007 in both Germanyand the Netherlands. A follow up survey was held among these nascententrepreneurs approximately one year after the adult population surveys.In this follow up survey it was assessed whether the nascent entrepreneurs

16

had succeeded in setting up a business, whether they were still in the processof setting up their firm or whether they abandoned their start-up attempt.This three-stage variable for success in firm foundation or ”start-up success”is the dependent variable in our models. Table 1 shows the numbers ofrespondents reached for each of these three categories.

Information on whether the business was founded is based on follow-up in-terviews that were conducted one year after the GEM-screening interviews.We cannot take account of the fact that during the first screening in whichthe nascent entrepreneurs were identified some start-up initiatives were al-ready more advanced than others.

Table 1: Status of the business start-up attempt one year after first screening

Number Percent

Business founded 109 57.7Postponed 40 21.2Given-up 40 21.2

Total 189 100.0

4.1.2 Innovation

We capture innovation by product (or service) innovations and process inno-vations. First, we distinguish between those who attempt to introduce newproducts or services to the markets and those who introduce already exist-ing goods or services. Second, we distinguish those who make use of newtechnologies from those who do not. While the former is about introducingproducts or services that are new to customers, the latter reflects using anew technology, which e.g. may make it possible to introduce somethingthat already exists at lower costs and thus can make it possible to attractnew, more price-sensitive customers (OECD 2005).

The measures that we use for product innovation and process innovation areboth based on questions contained in the GEM adult survey. Individualsidentified as nascent entrepreneurs are asked whether the planned businesswill offer a novel product or service or all, to some or to none of its (expected)customers. We create a dummy for product innovation getting the value 1in case an individual offers a product or service that is new to all or somecustomers. A process innovation is identified by the GEM adult population

17

survey population question regarding the age of the technology that is to beused. We construct a dummy variable that takes the value of one when theinterviewees use technologies that have been available for five years or lessand zero otherwise.

More than half of all nascent entrepreneurs in our sample claim that theservice or product they intend to supply is at least new to some customers,including 13 percent that even claim it new to all customers. These sharesare in line with results from the US PSED (Panel Study of EntrepreneurialDynamics) studies, where 19 percent of all nascent entrepreneurs in 2006state that no existing firms already offer (the) same product or (the) sameservice to the (expected) customers (Hurst and Pugsley, 2011: 94). Theshare of nascent entrepreneurs using new production technologies is notablysmaller than those with product or service innovations. Just a quarter ofall nascent entrepreneurs claim to use a technology that is younger than sixyears.

4.1.3 Skill variety and balance

Lazear (2005) chose a retrospective design by analyzing the employment-biographies of Stanford graduates via a job history panel to examine thevariety and the degree of specialization of each graduate. For him the mainmeasure of the number of skills of a person is the number of different occu-pations a person has had before. Lazear (2005) finds a positive relationshipbetween this measure and the probability for self-employment, citing this asevidence for the ‘jack-of-all-trades’ hypothesis (Lazear, 2004; 2005).

We analyze a sample of individuals attempting to establish a business. Herethe relevant question is not whether multi-skilled individuals are more likelyto become self-employed or business owners, but whether they are morelikely to succeed when trying to set up a business. This is a very differentquestion than that asked in previous studies, since the primary threshold,the decision to start a business, has already been made. Furthermore, oursample differs from that of Lazear (2005) in that it is not restricted tograduates of a university, but comes from a representative selection of theDutch and German population of nascent entrepreneurs.

It is certainly a challenge to assess the variety of skills a person masters andeven more so to judge how balanced these skills are. We did the former

18

by posing a question consisting of a list of eight items. The intervieweeswhere asked about how many years of experience they had in each of eightfields. These fields are: design/engineering, production, marketing/sales,finance/accounting, legal, human resources, general management and con-sultancy. We use this variable as a continuous variable for skill variety, butwe also constructed three dummies for low, medium and high skill vari-ety. If they had a least a full year of experience in a field, we rated thisas experienced (=1) and aggregated the nine items (0-8) into a tree-stageddummy-variable which we label ‘low skill variety’ in case one has experiencein less than three fields, ‘medium skill variety’ if one has experience in threeto five fields and ‘high skill variety’ in case the interviewee has experiencein six to eight fields.

In order to judge how balanced the nascent entrepreneurs’ skills are wechoose to rely on self-assessment. A personal assessment has the advantageover any approach to examine the balance of knowledge objectively in thatit considers the individual’s perception. This is important, since the indi-vidual’s perception of his or her own abilities is of great importance for thelikelihood to try to become self-employed (Simon et al., 2000). After posingthe questions about the fields of experience used for assessing the variety ofskills, a simple question was asked whether the interviewees would chooseto describe themselves overall more as being a generalist or more as being aspecialist in one specific area.

According to Lazear (2005) individuals who plan to start a new businessshould intentionally choose to acquire more varied skills than those who donot. Table 2 shows that 56 per cent of our sample claims to have experiencein 3 to 8 fields, while 44 percent has experience in 0 to 2 fields. Furthermore6 percent would describe themselves more as a generalist. Table 2 shows thatthose who claim to be generalists more often than specialists have experiencein some or many fields, and less often master only a few fields. Almost half(44 percent) of the specialists interviewed claims to have gained experiencein three or more different fields. Furthermore, more than one thirds of thegeneralists have experience in two fields or less only. The correlation betweenthe skill balance dummy and the continuous skill variety variable is as low as0.09, supporting our view that balance and variety of skills is not necessarilysimilar.

As a side-condition for his model Lazear (2005) stated that the model mightnot apply for solo-entrepreneurs since their firms often tend to be focused on

19

Table 2: Skill balance and skill variety

Number of fields of Skill balance: Skill balance: Skill varietyexperience Specialist Generalist

Few (0-2) 56 36 44Some (3-5) 35 47 43Many (6-8) 8 16 13

Total 100 (n=71) 100 (n=118) 100 (n=189)

a specific skill. We control for this by including a dummy variable indicatingwhether one is a solo-entrepreneur or not. Being a solo-entrepreneur is char-acterized by three criteria: not having a partner, not employing others (thatalso includes sub-contractors according to the rather strict GEM-definition)and not wishing to employ someone within the coming five years. Accordingto these criteria 13 percent of all nascent entrepreneurs in our sample canbe classified as solo-entrepreneurs.

4.2 Estimation Strategy and Results

Our estimation strategy is to test whether the variety (and balance) of skillsand the innovativeness of the planned start-up influence the probability tosuccessfully terminate the entrepreneurial process by establishing the busi-ness. To this end we use a multinomial logit model. The outcome of thestart-up attempt one year after the initial screening is the dependent vari-able, and can take the values of either being founded, postponed, or given-up. The independent variables are as described variables measuring skillbalance, skill variety, and including dummy variables for product innovationand process innovation. Additionally we include several variables to controlfor heterogeneity at the individual level, including age, sex, and education.A description of all the variables is given in Table 3.

We first test whether skill variety (and skill balance) matter for starting-upa new business successfully (Hypothesis 1). Table 4 includes the continuousskill variety variable and the skill balance variable and we find no significantrelationship between either of these measures and success in getting thebusiness founded. When we use the skill variety dummies (see Table 5),however, it can be seen that there is a significant and positive correlationbetween having a high skill variety (as opposed to a low skill variety) andstart-up success (p>0.05). This confirms Hypothesis 1. In addition we find

20

Table 3: Dependent, independent and control variables

Variable Description Reference Category/ Range

Dependent variableStart-up success Categorical:

- Business started in follow-up survey (1 year after first contract)- Start-up attempt postponed- Start-up attempt abandoned/given up

Independent variablesSkill variety Two measures (metric and dummies): Metric

- Number of fields (0-8) in which the respondent has at least oneyear experience- Three dummies based on the number of fields in which therespondent has at least one year experience: 0-2 (Low skill va-riety), 3-5 (Medium skill variety), 6-8 (High skill variety)

Low skill variety

Skill balance Dummy indicating whether the respondent considers himselfmore of a generalist as opposed to more of a specialist

Specialist

Product Innovation Dummy for introduction of products or services that are new toall or some customers

Product or service new tonone of the customers

Process innovation Dummy for age of production technology used. Production technology usedmore than 5 years old

Control variablesEducation Three dummies (low, medium and high) for level of education

of the respondentHigh education (universitydegree or higher)

Age, Age (2) Age of the respondent in years (squared) MetricWoman Dummy for gender of the respondent ManIndustry Experience (2) Number of months that the respondent has experience in the

same industry/area (ln), (squared)Metric

Recent employment Dummy for whether the respondent worked as an employee inthe recent past just before (attempting to) founding the business

Not employed before

Prior start-up experience Dummy for whether the respondent has prior experience withfounding a business

No prior start-up experience

Asked for advice Dummy indicating whether the respondent has sought profes-sional advice for establishing the business

No professional advicesought

Solo-entrepreneur Dummy indicating that the respondent does not have a businesspartner, does not employ others (this includes sub-contractors)and does not wish to employ someone within the coming fiveyears

Without business partner(s)and employees, and doesnot wish to employ someonewithin the next five years

High growth ambition Dummy indicating the growth expectations of the start-up interms of revenues

Below average growth ex-pectations

Investments > 10,000 Dummy for the respondents (intended) amount of investments (Intended) investments10,000 or lower

Industry Dummy for industry with the biggest share Real estate, renting andDummy for “industry not yet known” business activities

Country Dummy for the Netherlands GermanyYear of screening Dummy for 2007 2006

21

a significant negative relationship between high skill variety and abandoningor giving up the start-up suggesting that skill variety reduces such outcomes.One tentative explanation is that higher skill variety can lead to prospectiveentrepreneurs discovering their own talents, and hence discourage them fromespecially non-innovative start ups. The results in Tables 4 and 5 also revealthat being a generalist or having skill balance, which interestingly has anegative sign, does not seem to relate significantly to start-up success.

Considering the control variables we can note that there are a number of vari-ables that are significantly related to our dependent variable. Entrepreneurswith a medium level of education (as compared to those with a high level ofeducation) are more likely to succeed in getting their business founded. Thisfinding reflects the fact that better educated individuals have a wider rangeof alternative wage-employment options. Having siome recent experiencein wage-employment furthermore increases the probability of establishing abusiness. This may suggest that up-to-date employment experience equipindividuals with specific knowledge, experience, skills and networks that areuseful in the start-up process.

Intriguingly, it appears that those who sought professional advice are lesslikely to succeed in their start-up attempts. While we do not have a clear-cutexplanation for this it could be a sign of uncertainty or lack of knowledge orexperience when someone is looking for external advice. Alternatively, seek-ing professional advice could also be indicative for more complex start-upattempts and therefore result in a lower likelihood to succeed in establishinga business.

Finally, it can also be seen that some of the factors associated with an in-creased likelihood of being a business owner or self-employed (e.g. genderand having prior start-up experience) do not seem to matter for being suc-cessful in the start-up process.

In Tables 4 and 5 product and process innovation are included as controlvariables, given Hypothesis 2. It can be seen that innovation by itself doesnot seem to matter for start-up success. In the next tables we will includeinteraction terms for skill variety and skill balance on the one hand, andproduct and process innovation on the other hand, to evaluate whetherinnovation positively moderates the relationship of skill variety and skillbalance and start-up success.

22

In Table 6 the interaction term for skill variety and product innovation isincluded, while Table 7 includes the interaction term for skill variety andprocess innovation. In accordance with our expectations (Hypothesis 2) wefind support that product innovation, as well as process innovation, posi-tively moderate the relationship between skill variety and start-up success.

If skill variety differs from skill balance than the moderating effect of in-novation does not necessarily apply to skill balance. Table 8 and 9 indeedreveal that neither the coefficients/marginal effects of the interaction termsfor skill balance and product innovation nor for skill balance and processinnovation are significant.

Table 4: Multinomial logit results for start-up success with skill variety (metric)and skill balance included

Business Founded Postponed Given-updy/dx z P>|z| dy/dx z P>|z| dy/dx z P>|z|

Skill variety 0.02 1.18 0.238 0.00 −0.03 0.973 −0.02 −156 0.119Skill balance −0.11 −1.42 0.155 0.08 1.26 0.209 0.02 0.42 0.674Woman −0.00 −0.04 0.967 −0.04 −0.47 0.641 0.04 0.61 0.540Medium education 0.17 2.33 0.020 −0.12 −1.76 0.079 −0.06 −0.91 0.363Low education 0.04 0.41 0.681 −0.04 −0.45 0.651 0.00 0.03 0.974Age −0.02 −0.73 0.463 −0.01 −0.59 0.557 0.03 1.01 0.313Age2 0.00 1.15 0.248 0.00 0.43 0.664 0.00 −1.23 0.220Industry exp. (yrs) 0.02 1.48 0.138 0.01 1.02 0.307 −0.03 −2.80 0.005Industry exp.2 0.00 −1.92 0.055 0.00 −0.97 0.333 0.00 3.18 0.001Recent employment exp. 0.24 2.71 0.007 −0.21 −2.87 0.004 −0.03 −0.37 0.709Prior start-up exp. −0.06 −0.75 0.454 0.01 0.14 0.889 0.05 0.78 0.435Asked for advice −0.04 −2.59 0.010 0.03 1.63 0.102 0.02 1.31 0.189Solo-entrepreneur 0.03 0.26 0.793 −0.09 −0.88 0.376 0.05 0.42 0.671High growth ambition −0.08 −1.11 0.268 0.00 0.02 0.983 0.08 1.63 0.103Investments > 10,000 −0.05 −0.58 0.560 0.01 0.17 0.867 0.03 0.60 0.547

Notes: Industry, country and year dummy/dummies included, but not reported.

23

Table 5: Multinomial logit results for start-up success with skill variety dummiesand skill balance included

Business Founded Postponed Given-updy/dx z P>|z| dy/dx z P>|z| dy/dx z P>|z|

High skill variety 0.26 2.07 0.038 −0.01 −0.07 0.944 −0.25 −2.14 0.032Medium skill variety 0.03 0.37 0.712 −0.06 −0.91 0.365 0.03 0.59 0.558Skill balance −0.11 −1.45 0.147 0.09 1.31 0.189 0.02 0.38 0.701Medium education 0.18 2.50 0.013 −0.11 −1.76 0.079 −0.07 −1.10 0.270Low education 0.03 0.30 0.765 −0.04 −0.50 0.618 0.02 0.25 0.806Age −0.02 −0.87 0.382 −0.01 −0.50 0.618 0.03 1.13 0.257Age2 0.00 1.33 0.183 0.00 0.34 0.732 0.00 −1.39 0.165Woman −0.01 −0.08 0.939 −0.04 −0.59 0.555 0.05 0.85 0.394Industry exp. (yrs) 0.01 1.32 0.188 0.01 0.98 0.328 −0.02 −2.71 0.007Industry exp.2 0.00 −1.82 0.070 0.00 −0.95 0.344 0.00 3.22 0.001Recent employment exp. 0.24 2.77 0.006 −0.19 −2.66 0.008 −0.05 −0.67 0.501Prior start-up exp. −0.08 −1.04 0.297 0.02 0.31 0.757 0.06 1.03 0.302Asked for advice −0.04 −2.42 0.016 0.03 1.78 0.075 0.01 0.89 0.372Solo-entrepreneur 0.02 0.12 0.903 −0.09 −0.90 0.368 0.07 0.55 0.582High growth ambition −0.11 −1.47 0.142 0.00 −0.02 0.984 0.12 2.27 0.024Investments > 10,000 −0.04 −0.58 0.564 0.01 0.12 0.901 0.03 0.71 0.475

Notes: Industry, country and year dummy/dummies included, but not reported.

Table 6: Multinominal logit results for start-up success with interaction terms forskill variety (metric) and product innovation included

Business Founded Postponed Given-updy/dx z P>|z| dy/dx z P>|z| dy/dx z P>|z|

Skill variety −0.03 −1.42 0.154 0.03 1.08 0.280 0.00 0.18 0.860Skill balance −0.09 −1.24 0.216 0.07 1.05 0.293 0.02 0.38 0.700Product innovation −0.35 −3.13 0.002 0.19 1.64 0.102 0.15 1.59 0.112Process innovation −0.04 −0.53 0.597 −0.01 −0.12 0.906 0.05 0.89 0.372Skill variety x prod. in. 0.10 3.30 0.001 −0.06 −1.91 0.056 −0.04 −1.45 0.146Medium education 0.16 2.29 0.022 −0.11 −1.62 0.104 −0.05 −0.89 0.373Low education 0.03 0.30 0.761 −0.04 −0.40 0.690 0.01 0.12 0.904Age −0.03 −0.95 0.344 −0.01 −0.29 0.774 0.03 1.04 0.300Age2 0.00 1.32 0.187 0.00 0.16 0.875 0.00 −1.25 0.212Woman −0.01 −0.15 0.884 −0.03 −0.38 0.701 0.04 0.64 0.520Industry exp. (yrs) 0.02 1.55 0.121 0.01 0.96 0.338 −0.03 −2.95 0.003Industry exp.2 0.00 −1.95 0.051 0.00 −0.89 0.371 0.00 3.31 0.001Recent employment exp. 0.24 2.65 0.008 −0.21 −2.89 0.004 −0.02 −0.36 0.715Prior start-up exp. −0.06 −0.82 0.412 0.01 0.21 0.837 0.04 0.79 0.430Asked for advice −0.04 −2.59 0.010 0.03 1.65 0.098 0.02 1.18 0.237Solo-entrepreneur 0.02 0.15 0.882 −0.08 −0.78 0.437 0.06 0.46 0.644High growth ambition −0.06 −0.81 0.418 −0.02 −0.22 0.828 0.07 1.53 0.127Investments > 10,000 −0.04 −0.55 0.579 0.01 0.16 0.871 0.03 0.55 0.580

Notes: Industry, country and year dummy/dummies included, but not reported.

24

Table 7: Multinominal logit results for start-up success with an interaction termfor skill variety (metric) and process innovation included

Business Founded Postponed Given-updy/dx z P>|z| dy/dx z P>|z| dy/dx z P>|z|

Skill variety −0.01 −0.40 0.689 0.02 0.86 0.391 −0.01 −0.57 0.571Skill balance −0.10 −1.45 0.148 0.07 1.14 0.256 0.03 0.52 0.604Product innovation −0.10 −1.29 0.198 0.04 0.57 0.572 0.05 0.88 0.378Process innovation −0.41 −2.86 0.004 0.22 1.61 0.107 0.19 2.03 0.043Skill variety x proc. in. 0.12 3.18 0.001 −0.08 −2.06 0.040 −0.04 −1.62 0.105Medium education 0.17 2.41 0.016 −0.12 −1.75 0.079 −0.06 −0.95 0.344Low education 0.07 0.84 0.401 −0.06 −0.72 0.469 −0.01 −0.17 0.865Age −0.02 −0.87 0.386 −0.01 −0.35 0.727 0.03 0.98 0.327Age2 0.00 1.31 0.190 0.00 0.20 0.845 0.00 −1.21 0.225Woman −0.01 −0.08 0.935 −0.03 −0.40 0.687 0.04 0.58 0.560Industry exp. (yrs) 0.02 1.53 0.127 0.01 1.00 0.317 −0.03 −2.81 0.005Industry exp.2 0.00 −2.09 0.037 0.00 −0.81 0.417 0.00 3.24 0.001Recent employment exp. 0.25 2.85 0.004 −0.21 −2.97 0.003 −0.03 −0.47 0.636Prior start-up exp. −0.09 −1.29 0.196 0.03 0.43 0.671 0.06 1.08 0.278Asked for advice −0.04 −2.81 0.005 0.03 1.63 0.103 0.02 1.40 0.162Solo-entrepreneur 0.08 0.65 0.515 −0.13 −1.27 0.204 0.05 0.37 0.712High growth ambition −0.06 −0.80 0.424 −0.02 −0.22 0.826 0.07 1.52 0.130Investments > 10,000 −0.07 −0.95 0.342 0.03 0.43 0.667 0.04 0.74 0.458

Notes: Industry, country and year dummy/dummies included, but not reported.

Table 8: Multinomial regression results for start-up success with an interactionterm for skill balance and product innovation included

Business Founded Postponed Given-updy/dx z P>|z| dy/dx z P>|z| dy/dx z P>|z|

Skill variety 0.02 1.20 0.231 0.00 0.09 0.927 −0.02 −1.81 0.070Skill balance −0.24 −2.28 0.023 −0.04 −0.35 0.727 0.28 2.15 0.031Product innovation −0.20 −1.50 0.133 −0.10 −0.79 0.430 0.30 2.26 0.024Process innovation 0.00 −0.03 0.975 −0.04 −0.57 0.570 0.05 0.83 0.408Skill balance x prod. in. 0.18 1.20 0.230 0.16 1.23 0.217 −0.34 −2.47 0.013Medium education 0.18 2.38 0.017 −0.11 −1.73 0.084 −0.06 −1.06 0.290Low education 0.04 0.38 0.703 −0.04 −0.44 0.662 0.00 0.05 0.962Age −0.02 −0.83 0.407 −0.02 −0.79 0.430 0.04 1.18 0.240Age2 0.00 1.23 0.218 0.00 0.65 0.517 0.00 −1.37 0.172Woman −0.02 −0.21 0.835 −0.04 −0.54 0.592 0.06 0.98 0.328Industry exp. (yrs) 0.02 1.36 0.172 0.01 1.10 0.273 −0.03 −2.71 0.007Industry exp.2 0.00 −1.78 0.076 0.00 −1.12 0.262 0.00 2.99 0.003Recent employment exp. 0.26 2.78 0.005 −0.20 −2.81 0.005 −0.06 −0.82 0.414Prior start-up exp. −0.06 −0.74 0.459 0.01 0.12 0.908 0.05 0.84 0.401Asked for advice −0.05 −2.70 0.007 0.02 1.51 0.132 0.02 1.76 0.078Solo-entrepreneur 0.03 0.23 0.816 −0.10 −1.01 0.312 0.07 0.54 0.586High growth ambition −0.09 −1.19 0.236 0.00 −0.05 0.961 0.10 1.89 0.058Investments > 10,000 −0.05 −0.66 0.510 −0.01 −0.07 0.945 0.06 1.00 0.318

Notes: Industry, country and year dummy/dummies included, but not reported.

25

Table 9: Multinominal logit results for start-up success with an interaction termfor skill balance and process innovation included

Business Founded Postponed Given-updy/dx z P>|z| dy/dx z P>|z| dy/dx z P>|z|

Skill variety 0.02 1.18 0.239 0.00 0.01 0.995 −0.02 −1.57 0.117Skill balance −0.12 −1.49 0.135 0.06 0.83 0.404 0.06 0.83 0.408Product innovation −0.07 −0.90 0.367 0.02 0.29 0.774 0.05 0.78 0.435Process innovation −0.02 −0.11 0.911 −0.10 −0.72 0.470 0.12 1.21 0.228Skill balance x proc. in. 0.03 0.18 0.856 0.08 0.49 0.623 −0.11 −0.95 0.342Medium education 0.17 2.31 0.021 −0.12 −1.84 0.066 −0.05 −0.80 0.423Low education 0.04 0.40 0.688 −0.04 −0.48 0.634 0.01 0.07 0.943Age −0.02 −0.73 0.468 −0.01 −0.58 0.562 0.03 0.97 0.333Age2 0.00 1.13 0.529 0.00 0.43 0.667 0.00 −1.17 0.242Woman 0.00 −0.05 0.963 −0.03 −0.42 0.675 0.04 0.56 0.578Industry exp. (yrs) 0.02 1.44 0.149 0.01 1.03 0.303 −0.03 −2.72 0.007Industry exp.2 0.00 −1.87 0.062 0.00 −0.99 0.323 0.00 3.02 0.003Recent employment exp. 0.24 2.68 0.007 −0.21 −2.91 0.004 −0.03 −0.42 0.674Prior start-up exp. −0.05 −0.71 0.480 0.01 0.14 0.885 0.04 0.73 0.465Asked for advice −0.05 −2.64 0.008 0.02 1.58 0.115 0.02 1.48 0.138Solo-entrepreneur 0.03 0.23 0.820 −0.09 −0.90 0.367 0.06 0.49 0.622High growth ambition −0.09 −1.11 0.268 0.00 0.06 0.951 0.08 1.55 0.121Investments > 10,000 −0.05 −0.64 0.525 0.01 0.09 0.926 0.04 0.76 0.448

Notes: Industry, country and year dummy/dummies included, but not reported.

26

5 Concluding Remarks

In this paper we contributed in three ways towards extending Lazear’s the-ory of skills diversity and entrepreneurship. First, we explicitly linked en-trepreneurial skills to opportunities as an explanation for the rate of newbusiness creation. We proposed a theoretical model to show that opportuni-ties, skills and innovation interact to result in the creation of innovative newbusinesses. Second, we refined the ‘jack-of-all-trades’ concept to allow forgeneralists and specialists to possess a variety of skills through their experi-ences and/or investments in human capital. In other words, entrepreneurscan have a variety of skills even as they specialize. Our third contributionwas to test our model, in particular to provide empirical evidence on the in-teraction between an individual’s experience and skills on the one hand, andhis or her attempt to introduce a new product and/or service or a businessprocess innovation, on the other.

To test our model we use survey data collected among nascent entrepreneursin Germany and the Netherlands, supplemented by follow-up interviews oneyear after the first contact. Applying multinomial probit estimations, ourmain finding is that skill variety (high skill variety as opposed to low skillvariety) improves start-up success, while skill balance does not. Further-more it is found that prospective entrepreneur’s skill variety in particularfacilitates the establishment of innovative new business.

The main insights from this paper are as follows. Individuals with morevaried work experience are more likely to succeed in starting up a new busi-ness. Having experience in a variety of areas seems to facilitate the creationof a new business. This does not mean however that those with varied skillsare necessarily ‘more balanced individuals’ (Lazear, 2005: 676) as they maystill excel in one specific skill. As a matter of fact our empirical resultsshow that being a generalist, which proxies having balanced skills, does notmatter for success in setting up a business. This confirms our hypothesisthat having varied work experience does not necessarily imply that someoneis a specialist and vice versa. We find that this holds in particular for thoseattempting to set up an innovative business (i.e. introducing new productsor services to the market; and/or introducing new business processes).

Thus, innovation positively moderates the relationship between having var-ied experience and success in setting up a business. Those with skill varietyare better positioned to introduce innovations that not only have technical,

27

but also commercial value.

The policy implications from this paper are first that entrepreneurial edu-cation and business environment reform need to be tackled jointly and co-ordinated to improve matching between various skills and competencies onthe one hand, and opportunities on the other. Second, our findings supportthe notion that entrepreneurship can, at least partly, be learned, and thatlearning by doing and experiential learning, matters. These implicationssupports Cukier’s (2006:11) call for public policy toward entrepreneurshipto be more like ‘the very thing it hopes to promote- and embrace risk, exper-imentation and diversity’. Such an approach could broaden the experienceof prospective entrepreneurs. Finally, given that more varied work experi-ence is associated with start-up success, promotion of more flexible labormarkets and active labor policies that promote youth employment seem tobe supported by the model and evidence in this paper. Such institutionalreforms require however, as Lerner (2010) also stressed, persistence and along-term commitment of policy makers towards entrepreneurship.

Acknowledgements

We are grateful to Thomas Schøtt and Miguel Meuleman who providedvaluable input for the design of the questionnaire.

References

[1] Acs, Z.J. and Naude W.A. (2013). ‘Entrepreneurship, Stages of De-velopment, and Industrialization’, (In Szirmai, A., Naude, W.A. andAlcorta, L. (eds.). Pathways to Industrialization in the Twenty-FirstCentury. Oxford: Oxford University Press. Chapter 14, pp. 373-392.)

[2] Astebro, T. and Bernhardt. I. (2005). ‘The Winner’s Curse of HumanCapital’, Small Business Economics, 24: 63-78.

[3] Astebro, A., and Thompson, P. (2011). ‘Entrepreneurs, Jacks of alltrades or Hobos?’, Research Policy, 40: 637-649.

[4] Baptista, R., Karaoz, M. and Mendonca, J. (2007). ‘EntrepreneurialBackgrounds, Human Capital and Start-Up Success’, Jena EconomicResearch Papers 2007-045, Jena: Max Planck Institute of Economicsand Friedrich Schiller University.

28

[5] Braunerhjelm, P., Acs, Z.J., Audretsch, D.B., and Carlsson, B., (2010).‘The Missing Link: Knowledge Diffusion and Entrepreneurship in En-dogenous Growth,’ Small Business Economics, 34 (2): 105-125.

[6] Bruderl, J., and Schussler, R. (1990). ‘Organizational mortality: The li-abilities of newness and adolescence’, Administrative Science Quarterly,35: 530-547.

[7] Bruhn, M. and Zia, B. (2011). ‘Stimulating Managerial Capital inEmerging Markets: The Impact of Business and Financial Literacyfor Young Entrepreneurs’, Policy Research Working Paper no. 5642.Washington DC. The World Bank.

[8] Bublitz, E., and Noseleit, F. (2013). ‘The skill balancing act: whendoes broad expertise pay off?’ Small Business Economics, forthcoming,doi10.1007/s11187-013-9474-z

[9] Carter, N. M., Gartner, W. B., and Reynolds, P. D. (2004). ‘Firmfounding’. (In Gartner W. B., Shaver K. G., Carter N. M., and P.D. Reynolds (eds.). Handbook of Entrepreneurial Dynamics. ThousandOaks, London: Sage, pp. 311-323.)

[10] Cukier, K. (2006). Innovative Entrepreneurship and Public Policy: Herowith a Thousand Faces. Swiss Re.

[11] Evans, D.S. and Jovanovic, B. (1989). ‘An Estimated Model of En-trepreneurial Choice under Liquidity Constraints’, Journal of PoliticalEconomy, 97(4): 808-827.

[12] Freeman, J., Carroll, G.R., and Hannan, M.T. (1983). ‘The Liabilityof Newness: Age dependence in organizational death rates’, AmericanSociological Review, 48: 692-710.

[13] Gries, T., and Naude, W. (2011). ‘Entrepreneurship and human de-velopment - A capability approach’, Journal of Public Economics, 95:216-224.

[14] Hurst, E., and Pugsley, B.W. (2011). ‘What do small businesses do?’,Brookings Papers on Economic Activity, Economic Studies Program,The Brookings Institution, 43(2): 73-142.

[15] Karlan, D. and Valdivia, M. (2011). ‘Teaching Entrepreneurship: Im-pact of Business Training on Microfinance Clients and Institutions’ TheReview of Economics and Statistics, 93 (2): 510-527.

29

[16] Katz, J., and Gartner, W.B. (1988). ‘Properties of emerging organiza-tions’, Academy of Management Review, 13(3): 429-441.

[17] Kanniainen, V. and Poutvaara, P. (2007). ‘Imperfect Transmission ofTacit Knowledge and Other Barriers to Entrepreneurship’, ComparativeLabor Law and Policy Journal, 28: 675-693.

[18] Kihlstrom, R.E. and Laffont, J-J. (1979). ‘A General Equilibrium En-trepreneurial Theory of Firm Formation Based on Risk Aversion’, Jour-nal of Political Economy, 87(4): 719-748.

[19] Lazear, E.P. (2004). ‘Balanced skills and entrepreneurship’, AmericanEconomic Review, 94(2): 208-211.

[20] Lazear, E.P. (2005). ‘Entrepreneurship’, Journal of Labor Economics,23(4): 649-680.

[21] Lerner, J. (2010). ‘The Future of Public Efforts to Boost Entrepreneur-ship and Venture Capital’, Small Business Economics, 35: 255-264.

[22] Lucas, R.E. (1978). ‘On the Size Distribution of Business Firms’, BellJournal of Economics, 9: 508-523.

[23] OECD; Eurostat (2005, 3rd edition): Oslo Manual Guidelines for col-lecting and interpreting innovation data. OECD Publishing.

[24] Parker, S.C. (2009). The Economics of Entrepreneurship. Cambridge(MA): Cambridge University Press.

[25] Pissarides, C. A. (2000). Equilibrium Unemployment Theory. 2nd ed.:MIT Press.

[26] Reynolds, P.D., and Miller, B. (1992). ‘New firm gestation: Conception,birth, and implications for research’, Journal of Business Venturing, 7:1-8.

[27] Silva, O. (2007). ‘The Jack-of-All-Trades entrepreneur: Innate talentor acquired skill?’ Economics Letters, 97: 118-123.

[28] Simon, M., Houghton, S.M., Aquino, K. (2000). ‘Cognitive biases, riskperception, and venture formation: How individuals decide to startcompanies’, Journal of Business Venturing, 15(2): 113-134.

[29] Spulber, D.F. (2009). The Theory of the Firm: Microeconomics withEndogenous Entrepreneurs, Firms, Markets, and Organizations. Cam-bridge (MA): Cambridge University Press.

30

[30] Stuetzer, M., Goethner, M., and Cantner, U. (2012). ‘Do balanced skillshelp nascent entrepreneurs to make progress in the venture creationprocess?’, Economics Letters, 117: 186-188.

[31] Stuetzer, M., Obschonka, M., Davidsson, P., and Schmitt-Rodermund,E. (2013a). ‘Where do entrepreneurial skills come from?’ Applied Eco-nomics Letters, 20(12): 1183-1186.

[32] Stuetzer, M., Obschonka, M., and Schmitt-Rodermund, E. (2013b).‘Balanced skills among nascent entrepreneurs’, Small Business Eco-nomics, 41(1): 93-114.

[33] Toft-Kehler, R., Wennberg, K. and Kim, P.H. (2013). ‘Practice MakesPerfect: Entrepreneurial-Experience Curve and Venture Performance’,Journal of Business Venturing, Article in Press.

[34] Wagner, J. (2003). ‘Testing Lazear’s jack-of-all-trades view of en-trepreneurship with German micro data’, Applied Economic Letters,10, 687-689.

[35] Wagner, J. (2006). ‘Are nascent entrepreneurs ‘Jacks-of-all-trades’? Atest of Lazear’s theory of entrepreneurship with German data’, AppliedEconomics, 38(20): 2415-2419.

6 Appendices:

6.1 Appendix 1: Determining Optimal Effort Levels

Search effort: Determining the optimal search effort, the effort functionη, and the derivatives of η:

To simplify we may assume symmetry for the value of human capital in eachexperience hij . Hence (

∑eij=1 hij)

−1 = (eihi)−1.

maxηi

: rWi = wi −ηi

eihi+ µ

ωβ(ηiu)1−β

u∆i

First order conditions (FOCs) and second order conditions (SOCs):

0 = − 1

eihi+ µ

∆

u(1− β)

M

ηi, 0 > −µ∆i

u(1− β)βωβη−β−1

i u1−β

31

Optimal search effort of each entrepreneur is determined by using the im-plicit function theorem from the FOCs. and SOCs. We obtain

η∗i = η∗i (u, ω,∆i, ei, µ) , η∆i > 0, ηω > 0, ηu < 0, ηei < 0, ηµ > 0

Derivatives of the optimal effort:

η∆i =ηi

∆iβ> 0, ηω =

ηiω> 0, ηu = −ηi

u< 0,

ηei =1

µ∆iu (1− β)βM

η2i

1

eihi> 0, ηµ =

ηiµβ

> 0

Stay in market effort: Determining optimal effort to stay in the market,effort function ψi and derivatives:

maxψi

: rVi = vi − γiψi − φi (ψi) ∆i

FOCs and SOCs.

−γi − φψi∆i = 0, −φψiψi

∆i < 0

where φψi:= ∂φi

∂ψi. From the FOCs and SOCs we obtain the optimal strategy

ψ∗i = ψ∗i (∆i, γi)

with

∂ψi∂γi

=: ψγi = − 1

φψiψi∆i

< 0,∂ψi∂∆i

=: ψ∆ = −φψi

φψiψi∆i

> 0

6.2 Appendix 2: Proof of Proposition 1

Equations F,G,H [(11), (22), (33)] have continuous partial derivatives withrespect to all variables. As all variables are positive, and since

q

p+ qΩ− E > 0→ ω > u,

the determinant of the Jacobian matrix for the smooth function f(x, y) =(F,G,H)(x, y), y = (ω, u,∆) , x = (µ, e, q, p,Ω, E, w) does not vanish:

A =

−µMω −φ φψψ∆ (E − u)− µ (1− β) M∆β

∆µβmω −∆µβmu (r + φ+ µm)1 −1 0

32

|A| = −(r+φ+µm)

(φ+ µ

M

ω

)+µ

(−φψ

+ψ∆ (E − u) ∆β − µ (1− β)M

)(m

u− m

ω︸ ︷︷ ︸>0

) 6= 0

So that the Jacobian matrix is invertible and the implicit function theoremcan be applied. System [(11), (22), (33)] implicitly defines the functions

u∗ = u∗ (µ, e, h, q, p,Ω, E, v, w)

ω∗ = ω∗ (µ, e, h, q, p,Ω, E, v, w)

∆∗ = ∆∗ (µ, e, h, q, p,Ω, E, v, w) .

6.3 Appendix 3: Proof of Hypothesis 1 and 2

Comparative statics for the system F , G, H can be performed by taking thepartial reaction from A∂a = ∂B, with

∂a = (∂ω, ∂u, ∂∆)′,

A =

−µMω −φ−φψ

+ψ∆ (E − u)− µ (1− β) M

∆β

∆µβmω −∆µβmu (r + φ+ µm)1 −1 0

∂B =

Mβ ∂µ−

ηu∆β∂c

00

solving for the effects of the variety of work experiences e, and matchingefficiency µ on wealth differentials and the number of nascent entrepreneursgives for ∂∆∗

∂e and ∂∆∗

∂µ ,

∂∆∗

∂e=

− uη∆β

1e2h[(

µMω + φ) (r+φ+µm)

β∆µm( 1u− 1

ω )−(

(E − u)φψψ∆ − µ (1− β) M∆β

)] < 0,

∂∆∗

∂µ=

−Mβ[(

µMω + φ) (r+φ+µm)

β∆µm( 1u− 1

ω )−(

(E − u)φψψ∆ − µ (1− β) M∆β

)] < 0.

and for ∂u∗

∂e and ∂u∗

∂µ

33

∂u∗

∂e=

− uη∆β

1e2h[(

µMω + φ)−(

(E − u)φψψ∆ − µ (1− β) M∆β

)β∆µm( 1

u− 1

ω )(r+φ+µm)

] < 0,

∂u∗

∂µ=

−Mβ[(

µMω + φ)−(

(E − u)φψψ∆ − µ (1− β) M∆β

)β∆µm( 1

u− 1

ω )(r+φ+µm)

] < 0,

and for ∂ω∗

∂e and ∂ω∗

∂µ

∂u∗

∂e=∂ω∗

∂e,

∂u∗

∂µ=∂ω∗

∂µ.

a) Effects on start-up effort: Changes in search effort η are determinedby a number of variables η∗i = η∗i (u∗, ω∗,∆∗, ei, µ):

∂η∗i∂e

= −ηiu

∂u∗

∂e+ηiω

∂ω∗

∂e+

ηi∆β

∂∆∗

∂e+

1

µ∆u (1− β)βM

η2i

1

eih

= ηiηi

∆β

1

e2h

(r + φ)(1− u

ω

)[ (µMω + φ

)(r + φ+ µm)

−β∆µm(

1u −

1ω

) ((E − u)φψψ∆ − µ (1− β) M

∆β

) ] +1

µ∆u (1− β)βM

η2i

1

eih> 0

∂η∗i∂µ

= −ηiu

∂u∗

∂µ+ηiω

∂ω∗

∂µ+

ηi∆β

∂∆∗

∂µ+

ηiµβ

> 0

=ηiµ

uβ

(r + φ)(1− u

ω

)[ (µMω + φ

)(r + φ+ µm)

−β∆µm(

1u −

1ω

) ((E − u)φψψ∆ − µ (1− β) M

∆β

) ] +ηiµβ

> 0

b) Effects on the separation rate and stationary matching rate:Changes in the seperation rate φ:

φ = φ (ψ∗ (∆∗ (x) , γ)) , withφψi

< 0, φψiψi> 0,

ψγi < 0, ψ∆i > 0

34

hence from the reaction of ∆∗ we obtain:

∂φ∗

∂e=

<0∂φ

∂ψ∗

>0∂ψ∗

∂∆∗

<0∂∆

∂e> 0,

∂φ∗

∂µ=

<0∂φ

∂ψ∗

>0∂ψ∗

∂∆∗

<0∂∆

∂µ> 0.

Changes in the matching rate ε as the percentage of newly started firmsε = M

n

ε =Mn

Under stationary conditions (n = 0) M = φn and hence ε = φ. Therefore,

∂ε∗

∂e=

∂φ∗

∂e> 0,

∂ε∗

∂µ=

∂φ∗

∂µ> 0.

35